home *** CD-ROM | disk | FTP | other *** search
/ Chip 2011 November / CHIP_2011_11.iso / Programy / Narzedzia / Calibre / calibre-0.8.18.msi / file_280 / nikkei_sub_industry.recipe < prev    next >
Text File  |  2011-09-09  |  5KB  |  108 lines

  1.  
  2. __license__   = 'GPL v3'
  3. __copyright__ = '2010, Hiroshi Miura <miurahr@linux.com>'
  4. '''
  5. www.nikkei.com
  6. '''
  7.  
  8. import re
  9. from calibre.web.feeds.recipes import BasicNewsRecipe
  10. import mechanize
  11. from calibre.ptempfile import PersistentTemporaryFile
  12.  
  13.  
  14. class NikkeiNet_sub_industory(BasicNewsRecipe):
  15.     title           = u'\u65e5\u7d4c\u65b0\u805e\u96fb\u5b50\u7248(\u7523\u696d)'
  16.     __author__      = 'Hiroshi Miura'
  17.     description     = 'News and current market affairs from Japan'
  18.     cover_url       = 'http://parts.nikkei.com/parts/ds/images/common/logo_r1.svg'
  19.     masthead_url    = 'http://parts.nikkei.com/parts/ds/images/common/logo_r1.svg'
  20.     needs_subscription = True
  21.     oldest_article  = 2
  22.     max_articles_per_feed = 20
  23.     language        = 'ja'
  24.     remove_javascript = False
  25.     temp_files = []
  26.  
  27.     remove_tags_before = {'class':"cmn-section cmn-indent"}
  28.     remove_tags = [
  29.                        {'class':"JSID_basePageMove JSID_baseAsyncSubmit cmn-form_area JSID_optForm_utoken"},
  30.                        {'class':"cmn-article_keyword cmn-clearfix"},
  31.                        {'class':"cmn-print_headline cmn-clearfix"},
  32.                          ]
  33.     remove_tags_after = {'class':"cmn-pr_list"}
  34.  
  35.     feeds = [ (u'\u65e5\u7d4c\u4f01\u696d',     u'http://www.zou3.net/php/rss/nikkei2rss.php?head=sangyo'),
  36.          (u'\u65e5\u7d4c\u88fd\u54c1',    u'http://www.zou3.net/php/rss/nikkei2rss.php?head=newpro'),
  37.          (u'internet',        u'http://www.zou3.net/php/rss/nikkei2rss.php?head=internet'),
  38.          (u'\u56fd\u969b',         u'http://www.zou3.net/php/rss/nikkei2rss.php?head=kaigai'),
  39.          (u'\u79d1\u5b66',         u'http://www.zou3.net/php/rss/nikkei2rss.php?head=kagaku'),
  40.  
  41.         ]
  42.  
  43.     def get_browser(self):
  44.         br = BasicNewsRecipe.get_browser()
  45.  
  46.         cj = mechanize.LWPCookieJar()
  47.         br.set_cookiejar(cj)
  48.  
  49.         #br.set_debug_http(True)
  50.         #br.set_debug_redirects(True)
  51.         #br.set_debug_responses(True)
  52.  
  53.         if self.username is not None and self.password is not None:
  54.             #print "----------------------------get login form--------------------------------------------"
  55.             # open login form
  56.             br.open('https://id.nikkei.com/lounge/nl/base/LA0010.seam')
  57.             response = br.response()
  58.             #print "----------------------------get login form---------------------------------------------"
  59.             #print "----------------------------set login form---------------------------------------------"
  60.             # remove disabled input which brings error on mechanize
  61.             response.set_data(response.get_data().replace("<input id=\"j_id48\"", "<!-- "))
  62.             response.set_data(response.get_data().replace("gm_home_on.gif\" />", " -->"))
  63.             br.set_response(response)
  64.             br.select_form(name='LA0010Form01')
  65.             br['LA0010Form01:LA0010Email']   = self.username
  66.             br['LA0010Form01:LA0010Password'] = self.password
  67.             br.form.find_control(id='LA0010Form01:LA0010AutoLoginOn',type="checkbox").get(nr=0).selected = True
  68.             br.submit()
  69.             br.response()
  70.             #print "----------------------------send login form---------------------------------------------"
  71.             #print "----------------------------open news main page-----------------------------------------"
  72.             # open news site
  73.             br.open('http://www.nikkei.com/')
  74.             br.response()
  75.             #print "----------------------------www.nikkei.com BODY   --------------------------------------"
  76.             #print response2.get_data()
  77.             #print "-------------------------^^-got auto redirect form----^^--------------------------------"
  78.             # forced redirect in default
  79.             br.select_form(nr=0)
  80.             br.submit()
  81.             response3 = br.response()
  82.             # return some cookie which should be set by Javascript
  83.             #print response3.geturl()
  84.             raw = response3.get_data()
  85.             #print "---------------------------response to form --------------------------------------------"
  86.             # grab cookie from JS and set it
  87.             redirectflag = re.search(r"var checkValue = '(\d+)';", raw, re.M).group(1)
  88.             br.select_form(nr=0)
  89.  
  90.             self.temp_files.append(PersistentTemporaryFile('_fa.html'))
  91.             self.temp_files[-1].write("#LWP-Cookies-2.0\n")
  92.  
  93.             self.temp_files[-1].write("Set-Cookie3: Cookie-dummy=Cookie-value; domain=\".nikkei.com\"; path=\"/\"; path_spec; secure; expires=\"2029-12-21 05:07:59Z\"; version=0\n")
  94.             self.temp_files[-1].write("Set-Cookie3: redirectFlag="+redirectflag+"; domain=\".nikkei.com\"; path=\"/\"; path_spec; secure; expires=\"2029-12-21 05:07:59Z\"; version=0\n")
  95.             self.temp_files[-1].close()
  96.             cj.load(self.temp_files[-1].name)
  97.  
  98.             br.submit()
  99.  
  100.             #br.set_debug_http(False)
  101.             #br.set_debug_redirects(False)
  102.             #br.set_debug_responses(False)
  103.         return br
  104.  
  105.  
  106.  
  107.  
  108.